CN113221391A - Method for guiding material size parameter screening by analyzing circulator performance by BIC criterion - Google Patents
Method for guiding material size parameter screening by analyzing circulator performance by BIC criterion Download PDFInfo
- Publication number
- CN113221391A CN113221391A CN202110770107.4A CN202110770107A CN113221391A CN 113221391 A CN113221391 A CN 113221391A CN 202110770107 A CN202110770107 A CN 202110770107A CN 113221391 A CN113221391 A CN 113221391A
- Authority
- CN
- China
- Prior art keywords
- circulator
- parameter
- score
- value
- grades
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Artificial Intelligence (AREA)
- Pure & Applied Mathematics (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for guiding material size parameter screening by analyzing circulator performance by using BIC criterion, which comprises the steps of classifying circulators and determining a calculation mode of S parameter values according to classification; making a scoring rule, and quantifying each parameter; s parameter data of a circulator to be tested is obtained, and an overall score is obtained after quantification; and then, taking the size parameters of each material of the circulator as independent variables, taking the integral score as dependent variables, establishing a linear regression model under the BIC criterion, finding 1 or more independent variables which have obvious influence on the integral score, and guiding the material screening. The invention can take various material size parameters of the device as independent variables, take the quantitative score of a certain index or an integral index of the device as dependent variables to carry out regression analysis, and adopt the BIC criterion in statistics to find out a plurality of factors which have the most obvious influence on the quantitative score of the device performance, thereby guiding us to improve the material size parameters and improving the qualification rate of the device.
Description
Technical Field
The invention relates to a material screening method of a lumped parameter circulator, in particular to a method for guiding material size parameter screening by analyzing circulator performance by using a BIC (building information center) criterion.
Background
The lumped parameter circulator has a general structure comprising a shell, an internal central conductor braid assembly, three ports, a lumped capacitor, a load and the like, wherein the central conductor braid assembly comprises a central ferrite, a copper braid and the like. Different materials and parameters thereof have influence on product performance and qualification rate. In the early stage test of the batch production process of the traditional circulators/isolators, in order to improve the performance and the qualification rate of products, multiple test iterations are needed so as to determine the influence of different material size parameters on the performance and the qualification rate of the products. The traditional material screening process has the following defects:
(1) because the size of each actual material is necessarily a range rather than a theoretical value, it is difficult to find out the factors which are critical to the product performance by checking the change of the size parameter of each material through simulation software alone. How to find out the materials which do not need to be additionally enhanced in parameter precision control and how to excavate the material size factors which need to be further concerned is an urgent problem to be solved. In the traditional material screening method, products with qualified performance and unqualified products are found out through a plurality of rounds of tests and are compared, the difference of the sizes of the two types of products is checked, and the traditional material screening method can only carry out qualitative analysis generally.
(2) The evaluation of the performance and the qualification condition of the product is rough, generally speaking, the method adopted for judging whether the performance of the microwave device is qualified is that a tester directly measures by using a vector network analyzer to check whether each index in the working bandwidth range meets the index requirement. If the requirement is met, the product is qualified, otherwise, the product is unqualified. The judging method can only qualitatively judge whether one device is good or not, and can not judge whether the two devices are good or bad, so that the difference between the devices and the qualified standard is more difficult to describe. An explanation is given below of some of the terms used in the analysis method described later.
The noun explains:
(1) definition of BIC criterion
The BIC criterion is fully called Bayesian Information criterion, English Bayesian Information criteria is a regression model quality judgment standard in statistics, a likelihood function in statistics is utilized, the basic assumption of the criterion includes that random errors of variables obey normal distribution, and the criterion can be used for researching and selecting the importance of the variables in the variation of the dependent variables caused by a plurality of independent variables. BIC is defined as follows:
where RSS is the sum of the squares of the residuals of the regression fit of the data, p is the number of independent variables introduced into the model, and n is the sample volume.
The BIC criteria is a very effective method for selecting variables under a small sample high dimension.
(2) Screening principle of BIC criterion
As the number of variables introduced into the model increases,
the fitted residual sum of squares RSS in the equation will decrease, i.e. the first term will decrease, and
the second term in the equation will increase as the number of variables increases. When the effect of variance reduction brought by the increase of the variable is greater than the penalty brought by the increase of the variable, the value of the BIC is gradually reduced; when the number of the variables reaches a certain number, the penalty caused by the increase of the variables is larger than the variance reduction caused by the increase of the variables, and the value of the BIC is increased at the moment.
(3) Principles of BIC method selection
The model that minimizes BIC is an "optimal" model, which may include zero, one, or more of the arguments we have selected (i.e., material parameters), and when the result of the selection is zero, it indicates that all considered material parameters have no significant impact on device performance, otherwise, the selected argument is the material parameter that has a significant impact on device performance.
(4) With respect to regression models
For the product data in this document, we assume that there are 17 types of measurement parameters in the raw data, certainly not limited to these 17 types, and the measurement parameters can be increased or decreased according to the actual situation, for convenience of description, we make clear that there are 17 types here, which are sequentially denoted as X1 … … X17, and since there are many independent variables involved and the true physical relationship between the independent variables and the performance index (dependent variable) is extremely complex, we cannot describe them by analytical expressions, but we only analyze the influence of the fluctuation of each independent variable on the fluctuation of the device performance. Thus, a simple and easy assumption without loss of generality is that it conforms to the basic multiple linear regression model (the degree of linearity depends on the R in the result2Judgment):
in the above formula, β0Is intercept, beta1-β17Is an independent variable X1-X17The regression coefficient to be solved for is determined,
for the sum of random errors from the individual material tests, it is assumed that the random errors from the tests satisfy a normal distribution, which is generally satisfied during mass production.
According to the BIC criterion, in the basic regression model, we can select 0 independent variables (containing only intercept terms) and 1 independent variable … … 17 independent variables in turn to form the corresponding model, and the total number is
And (4) planting the model. The BIC criterion is that
And finding the model which minimizes the BIC quantity from the model types, thereby achieving the purpose of selecting the most main influence factors.
（5）R2Description of the invention
In multiple regression, R2Is defined as
Wherein
The average of the response variables, i.e., the dependent variable, i.e., the score herein,
the value of a response variable for a term,
is the value of the response variable under the corresponding independent variable predicted according to the regression model.
The statistical significance of (a) is to characterize how well the change in the dependent variable can be interpreted by the linear regression model, i.e., the linearity.
(6) Description of the value of p
In order to better understand the meaning of the p value of a certain coefficient in the subsequent R software result, the following explanation is made for the product influence factors: we have an original assumption that a factor has no significant relation with the product score, i.e. it is considered that the score difference is caused by sampling error, and an alternative assumption that the product score is significantly related to a factor. The significance level is set herein to 0.05 and the p-value is compared to the significance level. The meaning of p-value is the probability that the original hypothesis occurs in extreme cases, while small probability events are generally considered not to occur in one sample, so as long as p-value is less than 0.05, we consider it reasonable to reject the small probability event of the original hypothesis, so that a certain factor and performance score are considered to be significantly correlated, and the smaller the p-value, the more confident is that a decision is made to reject the original hypothesis for a certain factor. The smaller the corresponding independent variable p value in the R software calculation result is, the more remarkable the relationship between the independent variable and the dependent variable is.
(7) F statistic
The F statistic is a statistic used to test the significance of the regression model as a whole. When the F statistic of the entire model is greater than the F statistic at a given significance level, we reject the original assumption mentioned above and consider the model as a whole significant, i.e., the corresponding independent variables have a significant impact on the results.
The calculation of the parameters can be conveniently completed by using lm.bic functions in R statistical software, and the details of the calculation are not repeated.
Disclosure of Invention
The invention aims to provide a method for solving the problems, quantifying general indexes such as standing waves, isolation and loss, making a general scoring rule for device performance, finding out key factors influencing product performance, guiding material screening in the early stage, improving product quality and guiding material size parameter screening by analyzing circulator performance by using a BIC (building information center) criterion.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows: a method for guiding material size parameter screening by analyzing circulator performance by using BIC criterion comprises the following steps:
(1) dividing the circulators into A, B types, and determining the calculation mode of S parameter values;
the class A is a narrow band and is not more than 1 peak/valley in a working frequency band, and the rest is class B;
the S parameters comprise four parameters, namely a port standing wave ratio S11, an isolation degree S12, an insertion loss S21 and a two-port standing wave ratio S22;
if the circulator is a type A circulator, the S parameter value is the arithmetic mean value of the performance test values of the high-frequency point and the low-frequency point, and if the circulator is a type B circulator, the S parameter value is the interval mean value of the performance test values of the high-frequency point and the low-frequency point;
(2) making a scoring rule, and quantifying each parameter;
determining the qualified score and the full score of each parameter, wherein the full scores of the four parameters are 100 points in total;
for each parameter, defining S parameter values corresponding to 0 score, qualified score and full score, and quantizing the S parameter values corresponding to 0 to full score to obtain a quantized score;
(3) taking a circulator to be measured, obtaining S parameter data of S11, S12, S21 and S22 according to the step (1), obtaining four quantization fractions according to the step (2), and adding the four quantization fractions to obtain an overall score;
(4) and (2) establishing a BIC linear regression model by taking the size parameters of each material of the circulator as independent variables and the integral score as dependent variables, and finding 1 or more independent variables which have obvious influence on the integral score according to the BIC criterion.
Preferably, the method comprises the following steps: (5) and (5) guiding material screening according to the independent variable obtained in the step (4).
Preferably, the method comprises the following steps: the independent variables comprise weaving thickness, high-temperature pressing thickness, ferrite thickness, maximum enveloping diameter after weaving, one-port inductance line width, one-port gap, two-port inductance line width, two-port gap, three-port inductance line width, three-port gap, two-port and three-port weaving band angle, one-port and two-port weaving band angle, capacitor C1, capacitor C2, capacitor C3 and resistor R, and the respective variables are marked as X1-X17 in sequence.
Preferably, the method comprises the following steps: for the A-type circulator, P independent variables are included, the independent variables comprise ferrite thickness, maximum envelope diameter after weaving, inductance line width of each port, inductance clearance of each port and capacitance value of each port, and the independent variables are marked as X1-XP.
Preferably, the method comprises the following steps: the circulator is a lumped parameter circulator applied to a 5G base station, and the working frequency of the circulator is 1.805 GHz-1.88 GHz.
Preferably, the method comprises the following steps: the step (2) makes a scoring rule, and quantifies each parameter; the method specifically comprises the following steps:
the qualification grades of S11 and S22 are 25 grades and full grades are 30, the qualification grade of S12 is 15 grades and full grades is 20 grades, and the qualification grade of S21 is 14 grades, the full grades are 20 grades, and the total of four full grades is 100 grades;
s11 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
the S parameter value is corresponding to the standing-wave ratio;
s22 corresponds to a quantization value of
Method of calculation and
the same;
s12 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
the S parameter value corresponding to the isolation degree;
s21 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
is the value of the S parameter corresponding to the insertion loss. The insertion loss test value is actually a negative number, but the insertion loss test value is generally an absolute value in the industry, and the absolute value is substituted for the subsequent quantitative score value to calculate.
Regarding circulator classification: the lumped parameter circulator mainly aims at the lumped parameter circulator applied to a 5G base station, the size of the lumped parameter circulator is 5mm x 2.4mm, the working frequency of the lumped parameter circulator is 1.805 GHz-1.88 GHz, and the lumped parameter circulator is called as a 1.8 product, namely a class A circulator. The invention also uses a broadband circulator with the working frequency of 37.5GHz-41.5GHz as an embodiment of the B-class circulator.
Regarding the material size parameters: the type of circulator and the specific material used are determined according to the specific type of circulator, for example, the type a circulator has P =17 material size parameters, and the independent variables are 17, including the ferrite thickness, the maximum envelope diameter after weaving, the inductance line width of each port, the inductance gap of each port, and the capacitance value of each port, which may be specifically: the method comprises the following steps of knitting thickness, high-temperature pressing thickness, ferrite thickness, maximum envelope diameter after knitting, one-port inductance line width, one-port gap, two-port inductance line width, two-port gap, three-port inductance line width, three-port gap, two-port and three-port braid angle, one-port and two-port braid angle, capacitor C1, capacitor C2, capacitor C3 and resistor R, wherein variables of the three variables are marked as X1-X17 in sequence.
And the independent variables of the B-type circulator comprise ferrite length, ferrite width, ferrite height, T-shaped ceramic width, T-shaped ceramic length and the like.
With respect to arithmetic mean and interval mean, we illustrate two classes of circulators.
(1) Arithmetic mean: the average value of the performance test values of one device at a low frequency point and a high frequency point is obtained; referring specifically to fig. 3 a-3 d, fig. 3 a-3 d show the results of lumped parameter circulator testing of a 1.8 product numbered 1YZ-26#, which may be referred to as a narrowband device and also has only one peak/valley, i.e., a class a circulator. In the figure, the two-port standing wave ratio is taken as an example, the standing wave ratio at the low frequency point of 1.805GHz is 1.16, the standing wave ratio at the high frequency point is 1.24, and therefore the arithmetic average value of the standing wave ratios at the two ends is 1.20. Then the value of the S parameter of the circulator standing wave ratio is 1.20.
(2) Interval average value: the method is characterized in that integration is carried out between a low-frequency point and a high-frequency point by taking the frequency as an abscissa and taking a device S parameter (such as standing-wave ratio) as an ordinate, so that the area enclosed by a curve and an X axis is obtained, and the obtained result is divided by the length of the abscissa (namely frequency difference) to be used as the average value of the device performance in the interval. Referring to fig. 5, a performance test result of a wideband circulator in the range of its operating bandwidth (37.5 GHz to 41.5 GHz) is shown. It can be seen that the standing wave has a plurality of peaks/valleys. At the moment, the arithmetic mean value of the standing-wave ratios at two ends is not accurate only by calculating, original test data needs to be derived by a vector network analyzer, the standing-wave ratio on each frequency point in a working band is checked, then the area formed by a curve and an abscissa is obtained by numerical integration, and the obtained result is the interval mean value of the standing-wave ratios by dividing by the difference of the abscissa.
Compared with the prior art, the invention has the advantages that:
(1) provides a new method for guiding material screening. The method mainly comprises the following steps: the general indexes of standing wave, isolation, loss and the like are quantified firstly, a general scoring rule for the performance of the device is formulated, and each index of each device is qualitatively judged instead of being qualified or not. Then, various material size parameters of the device are used as independent variables, the quantitative score of a certain index or an integral index of the device is used as a dependent variable to carry out regression analysis, a general variable screening method in statistics, namely a BIC rule, is adopted to find out the factor which has the largest influence on the quantitative score of the device performance, more precisely, 1 or more independent variables with the most obvious influence can be given from large to small, and therefore materials which need to pay important attention in the material screening process are guided. For example, if the most significant independent variable is found to be the one-port and three-port braid angle, the "one-port and three-port braid angle" is specifically analyzed and improved, thereby improving the product yield.
(2) The method is suitable for material size guidance of the microwave device, which takes the S parameter as a main evaluation index of the device and is related to process step determination, device performance and various material size parameters. The S parameters comprise four parameters, namely a port standing wave ratio S11, an isolation degree S12, an insertion loss S21 and a two-port standing wave ratio S22; the material has more than five material size parameters.
(3) The invention carries out different data processing modes aiming at different types of circulators, thereby ensuring the accuracy of data.
(4) Before improvement, performance analysis and unqualified product analysis of the lumped parameter circulator are qualitative analysis, and principle reasons can be probably given out, but it is difficult to say that the specific unqualified material has no good control on which size, and the qualification rate fluctuates around 30% once.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a schematic diagram of a lumped parameter circulator of the present invention;
FIG. 3a is a graph of the results of the S11 test on a lumped parameter circulator for the 1.8 product numbered 1YZ-26# in example 2;
FIG. 3b is a graph of the results of the S21 test on a lumped parameter circulator for the 1.8 product numbered 1YZ-26# in example 2;
FIG. 3c is a graph of the results of the S12 test on the lumped parameter circulator for the 1.8 product numbered 1YZ-26# in example 2;
FIG. 3d is a graph of the results of the S22 test on the lumped parameter circulator for the 1.8 product numbered 1YZ-26# in example 2;
FIG. 4a is a graph of the S11 test results for the lumped parameter circulator for the 1.8 product numbered 1SY4-2# in example 2;
FIG. 4b is a graph of the S21 test results for the lumped parameter circulator for the 1.8 product numbered 1SY4-2# in example 2;
FIG. 4c is a graph of the S12 test results for the lumped parameter circulator for the 1.8 product numbered 1SY4-2# in example 2;
FIG. 4d is a graph of the S22 test results for the lumped parameter circulator for the 1.8 product numbered 1SY4-2# in example 2;
FIG. 5a is a graph showing the result of the S11 test on the standing-wave ratio of one port of the circulator in example 4;
FIG. 5b is a graph showing the results of the test of the isolation S12 of the circulator in example 4;
fig. 5c is a graph showing the test result of the insertion loss S21 of the circulator in example 4;
FIG. 5d is a graph showing the result of the S22 test on the standing-wave ratio of two ports of the circulator in example 4;
fig. 6 is a schematic structural diagram of a 37.5GHz-41.5GHz wideband circulator according to embodiment 4.
In the figure: 1. a housing; 2. a central ferrite; 3. a port; 4. two ports; 5. three ports; 6. a load; 7. the width of the conduction band; 8. a conduction band gap; 9. copper braided strap.
Detailed Description
The invention will be further explained with reference to the drawings.
Example 1: referring to fig. 1 and 2, fig. 2 shows a structural schematic diagram of a specific lumped parameter circulator, which includes a housing 1, a central ferrite, a port 3, a port 4, three ports 5, a load 6, a conduction band, a copper braid 9, etc., wherein the conduction band has indexes of conduction band width 7 and conduction band gap 8, the load 6 is placed at three fractures, the structure is actually an isolator, and the 3 ports have lumped parameter capacitors C1, C2, and C3, microwave signals enter from the port 3, are output from the port 4, and are in an isolated state from the port 4 to the port 3.
Based on the structure, a method for guiding material size parameter screening by analyzing circulator performance by using a BIC criterion is provided, which comprises the following steps:
(1) dividing the circulators into A, B types, and determining the calculation mode of S parameter values;
the class A is a narrow band and is not more than 1 peak/valley in a working frequency band, and the rest is class B;
the S parameters comprise four parameters, namely a port standing wave ratio S11, an isolation degree S12, an insertion loss S21 and a two-port standing wave ratio S22;
if the circulator is a type A circulator, the S parameter value is the arithmetic mean value of the performance test values of the high-frequency point and the low-frequency point, and if the circulator is a type B circulator, the S parameter value is the interval mean value of the performance test values of the high-frequency point and the low-frequency point.
(2) Making a scoring rule, and quantifying each parameter;
determining the qualified score and the full score of each parameter, wherein the full scores of the four parameters are 100 points in total;
for each parameter, defining S parameter values corresponding to 0 score, qualified score and full score, and quantizing the S parameter values corresponding to 0 to full score to obtain a quantized score; the method specifically comprises the following steps:
the qualification grades of S11 and S22 are 25 grades and full grades are 30, the qualification grade of S12 is 15 grades and full grades is 20 grades, and the qualification grade of S21 is 14 grades, the full grades are 20 grades, and the total of four full grades is 100 grades;
s11 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
the S parameter value is corresponding to the standing-wave ratio;
s22 corresponds to a quantization value of
Method of calculation and
the same;
s12 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
the S parameter value corresponding to the isolation degree;
s21 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
is the value of the S parameter corresponding to the insertion loss.
(3) And (3) taking a circulator to be measured, obtaining S parameter data of S11, S12, S21 and S22 according to the step (1), obtaining four quantization fractions according to the step (2), and adding the four quantization fractions to obtain an overall score.
(4) Taking the size parameters of each material of the circulator as independent variables, taking the integral score as dependent variables, establishing a BIC linear regression model, and finding 1 or more independent variables which have obvious influence on the integral score according to the BIC criterion; for a class a circulator, its independent variables include ferrite thickness, maximum envelope diameter after weaving, inductance line width of each port, inductance gap of each port, and capacitance value of each port, which are P in total, and each variable is marked as X1-XP, in this embodiment, we specifically refer to: the independent variables comprise weaving thickness, high-temperature pressing thickness, ferrite thickness, maximum enveloping diameter after weaving, one-port inductance line width, one-port gap, two-port inductance line width, two-port gap, three-port inductance line width, three-port gap, two-port and three-port weaving band angle, one-port and two-port weaving band angle, capacitor C1, capacitor C2, capacitor C3 and resistor R, and the respective variables are marked as X1-X17 in sequence.
In the invention, the circulator is a lumped parameter circulator applied to a 5G base station, and the working frequency of the circulator is 1.805 GHz-1.88 GHz. Although not limited thereto.
Example 2: based on embodiment 1, we provide a specific method for guiding material size parameter screening by analyzing circulator performance by using BIC criterion for a class A circulator.
Preparing two sets of circulators with working frequencies of 1.805 GHz-1.88 GHz, wherein the working frequencies are respectively as follows:
batch 1: numbers 1YZ-1# -1 YZ-50#, which are batches to which the devices of FIGS. 3 a-3 d belong;
batch 2: numbers 1SY4-1# -1SY4-42#, which are batches to which the devices in FIG. 4 a-FIG. 4d belong;
the two circulators are A-type circulators, the low frequency point is 1.805GHz, and the high frequency point is 1.88 GHz.
(1) Determining the type A circulator, wherein the S parameter value of the type A circulator is the arithmetic average value of the performance test values of the high frequency point and the low frequency point;
(2) making a scoring rule, and quantifying S11, S12, S21 and S22 in sequence according to the scoring rule of the embodiment 1;
(21) quantization of S11:
in this embodiment, the full score of S11 is 30 minutes, the standing waves at both ends in the operating band are both smaller than 1.2, which is defined as the full score, the standing waves at both ends in the normal temperature band are both smaller than 1.3, the return loss expressed by dB is greater than 17.69dB, which is qualified, and the qualified corresponding score is 25 minutes. When a scoring rule is formulated, the basic principle is that the farther the score deviates from an ideal value, the faster the score corresponding to the same value of standing wave ratio reduction is reduced, the score reduction speed corresponding to the same standing wave ratio range is consistent, namely, the standing wave average value and the corresponding score are in a piecewise linear function relationship, and the other three parameters are also based on the principle. According to the method of calculating S11 in the examples, we can obtain table 1:
table 1: port standing wave ratio S11 and quantization score comparison table
In the microwave device, the conversion relation between the return loss RL and the standing wave ratio rho is as follows:
order to
The S parameter value is corresponding to the standing-wave ratio; the table above is written as a piecewise function:
in this embodiment, since the circulator is a class a circulator
The arithmetic mean of the performance test values of the high frequency point and the low frequency point is adopted.
(22) S22 is quantified, and the method is identical to (21).
(23) Quantifying the isolation S12;
the normal temperature two-end isolation lower than-20 dB is defined as 20 minutes of full score, the two-end isolation lower than-17 dB is defined as qualified, the qualified is defined as 15 minutes, and the specific two-end isolation corresponding fraction is shown in the following table 2.
TABLE 2 isolation and quantification
Order to
The average isolation values of the low frequency point and the high frequency point in the band are represented, and the relation between the score and the average isolation value in the table can be represented as a piecewise function
(34) Quantifying insertion loss S21
The product with the loss at two ends of 1.8 at the normal temperature is lower than 0.55 and is defined as qualified, the corresponding score of the qualified product is 14 points, the loss at two ends is lower than 0.4 and is defined as full score of 20 points, and when the loss at two ends is higher than 0.55dB, the product is deducted for 3 points every time 0.05dB is increased. Table 3 was obtained.
Table 3: comparison table of insertion loss and score
Order to
The average values of the low frequency point and the high frequency point are shown, and S21 loss and corresponding score sum can be obtained according to the table
The functional relationship of (a) is:
(3) and (3) sequentially calculating S parameter data of a batch of circulators to be measured, obtaining four quantization divisions according to the step (2), and adding the four quantization divisions to obtain an overall score.
For batch 1 product, we describe by way of example a circulator numbered 1YZ-3 #:
fig. 3 a-3 d show the results of the lumped parameter circulator test for the 1.8 product numbered 1YZ-26#, four graphs being a port standing wave S11, insertion loss S21, isolation S12, and a port standing wave S22. Taking the two-port standing wave ratio as an example, the standing wave ratio at the low frequency point of 1.805GHz is 1.16, and the standing wave ratio at the high frequency point is 1.24, so the arithmetic mean value of the standing wave ratios at the two ends is 1.20. Then the value of the S parameter of the circulator standing wave ratio is 1.20. Similarly, the values of the S parameter corresponding to the other three parameters can be obtained, which will be described in detail below.
(31) S11: the standing wave ratio at the low frequency point of 1.805GHz is 1.37, the standing wave ratio at the high frequency point is 1.21, so that the arithmetic mean value of the standing wave ratios at the two ends is 1.29, that is, the S parameter value of S11 is 1.29, and according to the piecewise function calculation after table 1, it can be known that a port standing wave S11 has a score of 25.5, that is, a quantized score of 25.5.
(32) S12: the isolation at 1.805GHz at the low frequency point is-14.29 and the isolation at the high frequency point is-18.95, so the arithmetic mean of the isolation at both ends is-16.62. The isolation score of the product is 14.6 points by calculation according to the piecewise function shown in table 2.
(33) S21, the insertion loss at the low frequency point of 1.805GHz is-0.60, the insertion loss at the high frequency point of 1.88GHzA is-0.58, the arithmetic mean value of the two points is-0.59, the calculation is carried out according to the piecewise function after the table 3, and the calculation is carried out by substituting the absolute value of 0.59, so that the score of the insertion loss of the product is 11.6.
(34) S22, the standing wave ratio at the low frequency point of 1.805GHz is 1.16, the standing wave ratio at the high frequency point is 1.24, so the arithmetic mean value of the standing wave ratios at the two ends is 1.20. Referring to table 1, the two-port standing wave score is 30.
For batch 2 product we describe the circulator with number 1SY4-2#, as an example:
the lumped parameter circulator test results for the 1.8 product numbered 1SY4-2# in fig. 4 a-4 d have the same distribution as in fig. 3 a-3 d. According to the method for calculating the S parameter values of S11, S12, S21 and S22 in the batch 1 product, the S parameter values of S11, S12, S21 and S22 of the 1.8 product with the number of 1SY4-2# in the batch 2 product are calculated respectively.
Finally, we can get the overall score of each product in two batches according to the above method.
(4) And (2) establishing a BIC linear regression model by taking the material size parameters of the circulator as independent variables and the integral score as dependent variables, and finding out 1 or more independent variables which have obvious influence on the integral score according to a BIC criterion. Specifically included are (41) to (43).
(41) Before building the BIC linear regression model, we need to determine independent variables, which in this embodiment specifically include a knitting thickness, a high-temperature pressing thickness, a ferrite thickness, a maximum envelope diameter after knitting, a port inductance line width, a port gap, a two-port inductance line width, a two-port gap, a three-port inductance line width, a three-port gap, a two-port and three-port angle, a one-port and two-port angle, a capacitor C1, a capacitor C2, a capacitor C3 and a resistor R, and the respective variables are sequentially labeled as X1-X17. This can be shown by table 4 below:
table 4 variable codes and their corresponding material size parameters
(42) The product of batch 1 was analyzed to find 1 or more independent variables that significantly affected the overall score. The overall score of each product in batch 1 has been calculated according to step (3), for each product in batch 1, taking X1-X17 as independent variables and the overall score as dependent variables, building a BIC linear regression model, and finding 1 or more independent variables having a significant influence on the overall score according to the BIC criteria. After batch 1 was analyzed, 1 or more independent variables that had a significant effect on the overall score were found to be X12. Only one significant influencing factor after screening.
(43) The product of batch 2 was analyzed by the method of step (42) to find 1 or more independent variables X16 and X12 that had a significant effect on the overall score.
Example 3: on the basis of embodiments 1 and 2, a specific method for guiding material screening by performing performance analysis on a lumped parameter circulator by applying a BIC (binary analysis) rule is provided for a class A circulator.
In this embodiment, a batch of circulators with a working frequency of 1.805GHz to 1.88GHz are prepared, where the low frequency point is 1.805GHz and the high frequency point is 1.88GHz, specifically:
batch 3: numbers of 1SY2-51# -1SY 2-93 #. And the method comprises the following steps:
(1) determining the A-type circulator, wherein the calculation mode of the S parameter value is an arithmetic mean value;
(2) a scoring rule is formulated, and according to the scoring rule of the embodiment 1, a port standing-wave ratio S11 is quantified firstly in the same way as the embodiment 1 and the embodiment 2.
(3) In calculation batch 3: the integral score of each product in the number 1SY2-51# -1SY 2-93#, the batch of products belongs to A-class circulators, and therefore each S parameter value is calculated by adopting an arithmetic mean value;
(4) the same as in example 1 and example 2. Find 1 or more independent variables that have a significant impact on the overall score as X12, X10, X15.
Results of example 2 and example 3:
from the results of example 2, example 3, we show by the following table 5:
TABLE 5 summary of significant influence factors of different batches
As can be seen from the table, X12 is always one of the most significant independent variables, and therefore, it is necessary to take a careful control of it. We need to perform step (5): and (5) guiding material screening according to the independent variable obtained in the step (4), specifically:
the qualification rates of the products in the several batches in a certain interval around the average value of X12 are counted by taking X12 as an examination target, and the conditions are summarized as shown in the following table 6:
TABLE 6 different batches
Near yield
In the context of Table 6, the following examples are,
: the average value of X12 is shown.
It can be seen that after X12 is well controlled, the device yield is significantly improved. It is noted here that analysis of the first two batches in the table has shifted our focus to X12, and we have found that X12 was actually in the first two batches
The near qualification rate is highest, and the theoretical X12 optimal value sum on the simulation model
This did not coincide, which prompted us to investigate the effects of the material itself. Finally, the material of a certain part is replaced. Analysis performed after the material change showed that X12 was still the most significant factor and that the optimal center value of X12 coincided with the theoretical value. Subsequently, on the premise of keeping the size parameter range of other materials unchanged, the size parameter value of the material X12 is screened in advance and selected
Plus or minus 1 percent, thereby greatly improving the qualification rate.
The statistical batches 1SY4-1# -1SY4-42# in the table are
The +/-1% qualification rate is 85.7%, the qualification rate is slightly reduced due to a slight new change in the process after the manual operation of a plurality of process steps is changed into the automatic operation of a machine in the later period, but the whole process still can be kept above 70%, and the improvement of the qualification rate is remarkable compared with the initial qualification rate of 20-30%.
To summarize:
in example 2, example 3, multiple rounds of data analysis consistently determined that the major material size affecting the overall performance of the lumped parameter circulator was X12.
According to the results, the materials of the corresponding parts are changed, and the control and screening of X12 are enhanced, so that the later-stage overall yield of the device is increased to about 70%.
The method can also iterate continuously, and the most critical factors influencing the performance are found out in the residual material sizes after one material size is eliminated.
Example 4: referring to fig. 5 a-5B, and fig. 6, based on embodiments 1 and 2, we provide a specific method for guiding material size parameter screening by analyzing circulator performance using BIC criteria for a class B circulator.
In this embodiment, a batch of circulators with operating frequencies of 37.5GHz to 41.5GHz are prepared, the low frequency point is 37.5GHz, the high frequency point is 41.5GHz, and the main structure thereof is shown in fig. 6, and the specific implementation is as follows:
batch 4: code KSB1-KSB 14. And the method comprises the following steps:
(1) determining a B-class circulator, wherein the calculation mode of S parameter values is interval average values;
(2) the performance index of the wide-band circulator selected by us is completely consistent with that of the class a circulator in the embodiment 1 by making a scoring rule, so that the scoring rules of the two circulations are also completely consistent. According to the scoring rule of example 1, a port standing wave ratio S11 is quantified in the same manner as in examples 1 and 2.
(3) In calculation batch 4: the integral score of each product in the numbers KSB1-KSB14, the batch of products belongs to a B-class circulator, so that each S parameter value is calculated by adopting an interval average value;
(4) and (3) processing the material size parameters and the integral scores by using a BIC (building information center) rule, and screening out variables which have the most obvious influence on the performance, wherein the specific method is the same as that of the embodiment 1 and the embodiment 2.
TABLE 7 dimension code and meaning of each material of 37.5GHz-41.5GHz B-type circulator
Referring specifically to fig. 6, for a clearer description, the length, width, etc. are identified in the figure by the english alphabet. In fig. 6, the ferrite length is L, the ferrite width is W, the ferrite height is H, T, the ceramic width 1 is W1, the T-shaped ceramic width 2 is W2, the T-shaped ceramic length 1 is L1, and the T-shaped ceramic length 2 is L2.
Through the screening of the BIC criterion, the obtained significant variables are as follows from big to small in sequence: x1: ferrite length, X2: ferrite width, T-shaped ceramic width 2.
(5) And (4) processing a result: and (5) guiding material screening according to the independent variable obtained in the step (4). The two most significant variables found here: the length and width of the ferrite accord with the expectation of theoretical understanding and simulation results of the structure, and the overall performance of the device is obviously improved after size comparison and control are carried out on the ferrite obtained by selecting different processing methods for processing the ferrite. The product quantity is small, so that the qualification rate is not counted, and the improvement of the overall performance is emphasized.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (5)
1. A method for guiding material size parameter screening by analyzing circulator performance by using BIC criterion is characterized by comprising the following steps: the method comprises the following steps:
(1) dividing the circulators into A, B types, and determining the calculation mode of S parameter values;
the class A is a narrow band and is not more than 1 peak/valley in a working frequency band, and the rest is class B;
the S parameters comprise four parameters, namely a port standing wave ratio S11, an isolation degree S12, an insertion loss S21 and a two-port standing wave ratio S22;
if the circulator is a type A circulator, the S parameter value is the arithmetic mean value of the performance test values of the high-frequency point and the low-frequency point, and if the circulator is a type B circulator, the S parameter value is the interval mean value of the performance test values of the high-frequency point and the low-frequency point;
(2) making a scoring rule, and quantifying each parameter;
determining the qualified score and the full score of each parameter, wherein the full scores of the four parameters are 100 points in total;
for each parameter, defining S parameter values corresponding to 0 score, qualified score and full score, and quantizing the S parameter values corresponding to 0 to full score to obtain a quantized score;
(3) taking a circulator to be measured, obtaining S parameter data of S11, S12, S21 and S22 according to the step (1), obtaining four quantization fractions according to the step (2), and adding the four quantization fractions to obtain an overall score;
(4) determining material size parameters of a circulator, taking the material size parameters as independent variables, taking the integral score as dependent variables, establishing a BIC linear regression model, and finding out 1 or more independent variables which have obvious influence on the integral score according to a BIC criterion.
2. The method of claim 1 for guiding material dimensional parameter screening using BIC criteria analysis circulator wherein: and (5) guiding material screening according to the independent variable obtained in the step (4).
3. The method of claim 1 for guiding material dimensional parameter screening using BIC criteria analysis circulator wherein: for the A-type circulator, P independent variables are included, the independent variables comprise ferrite thickness, maximum envelope diameter after weaving, inductance line width of each port, inductance clearance of each port and capacitance value of each port, and the independent variables are marked as X1-XP.
4. The method of claim 1 for guiding material dimensional parameter screening using BIC criteria analysis circulator wherein: the circulator is a lumped parameter circulator applied to a 5G base station, and the working frequency of the circulator is 1.805 GHz-1.88 GHz.
5. The method of claim 1 for guiding material dimensional parameter screening using BIC criteria analysis circulator wherein:
the step (2) makes a scoring rule, and quantifies each parameter; the method specifically comprises the following steps:
the qualification grades of S11 and S22 are 25 grades and full grades are 30, the qualification grade of S12 is 15 grades and full grades is 20 grades, and the qualification grade of S21 is 14 grades, the full grades are 20 grades, and the total of four full grades is 100 grades;
s11 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
the S parameter value is corresponding to the standing-wave ratio;
s22 corresponds to a quantization value of
Method of calculation and
the same;
s12 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
the S parameter value corresponding to the isolation degree;
s21 corresponds to a quantization value of
：
In the formula (I), the compound is shown in the specification,
is the value of the S parameter corresponding to the insertion loss.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110770107.4A CN113221391B (en) | 2021-07-08 | 2021-07-08 | Method for guiding material size parameter screening by analyzing circulator performance by BIC criterion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110770107.4A CN113221391B (en) | 2021-07-08 | 2021-07-08 | Method for guiding material size parameter screening by analyzing circulator performance by BIC criterion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113221391A true CN113221391A (en) | 2021-08-06 |
| CN113221391B CN113221391B (en) | 2021-09-07 |
Family
ID=77081148
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110770107.4A Active CN113221391B (en) | 2021-07-08 | 2021-07-08 | Method for guiding material size parameter screening by analyzing circulator performance by BIC criterion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113221391B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171602A (en) * | 2017-06-28 | 2018-06-15 | 永辉青禾商业保理(重庆)有限公司 | The method for building up and device of accounts receivable factoring business risk model |
| US20190178720A1 (en) * | 2017-12-08 | 2019-06-13 | Duke University | Imaging devices including dielectric metamaterial absorbers and related methods |
| US20190246927A1 (en) * | 2018-02-14 | 2019-08-15 | Cerenion Oy | Apparatus and method for electroencephalographic measurement |
| US20190265574A1 (en) * | 2016-06-22 | 2019-08-29 | Massachusetts Institute Of Technology | Methods and systems for optical beam steering |
| US20200143278A1 (en) * | 2014-09-11 | 2020-05-07 | Berg Llc | Bayesian causal relationship network models for healthcare diagnosis and treatment based on patient data |
| CN111967799A (en) * | 2020-09-08 | 2020-11-20 | 中国电子科技集团公司第九研究所 | Method for identifying materials and process problems of integrated parameter table-attached circulator |
| WO2020247949A1 (en) * | 2019-06-07 | 2020-12-10 | The Regents Of The University Of California | General form of the tree alternating optimization (tao) for learning decision trees |
| CN112886165A (en) * | 2021-02-26 | 2021-06-01 | 中国电子科技集团公司第九研究所 | Semi-open cavity type microstrip circulator/isolator assembly |
| US20210191272A1 (en) * | 2019-12-21 | 2021-06-24 | Qoniac Gmbh | Wafer Exposure Method Using Wafer Models and Wafer Fabrication Assembly |
-
2021
- 2021-07-08 CN CN202110770107.4A patent/CN113221391B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200143278A1 (en) * | 2014-09-11 | 2020-05-07 | Berg Llc | Bayesian causal relationship network models for healthcare diagnosis and treatment based on patient data |
| US20190265574A1 (en) * | 2016-06-22 | 2019-08-29 | Massachusetts Institute Of Technology | Methods and systems for optical beam steering |
| CN108171602A (en) * | 2017-06-28 | 2018-06-15 | 永辉青禾商业保理(重庆)有限公司 | The method for building up and device of accounts receivable factoring business risk model |
| US20190178720A1 (en) * | 2017-12-08 | 2019-06-13 | Duke University | Imaging devices including dielectric metamaterial absorbers and related methods |
| US20190246927A1 (en) * | 2018-02-14 | 2019-08-15 | Cerenion Oy | Apparatus and method for electroencephalographic measurement |
| WO2020247949A1 (en) * | 2019-06-07 | 2020-12-10 | The Regents Of The University Of California | General form of the tree alternating optimization (tao) for learning decision trees |
| US20210191272A1 (en) * | 2019-12-21 | 2021-06-24 | Qoniac Gmbh | Wafer Exposure Method Using Wafer Models and Wafer Fabrication Assembly |
| CN111967799A (en) * | 2020-09-08 | 2020-11-20 | 中国电子科技集团公司第九研究所 | Method for identifying materials and process problems of integrated parameter table-attached circulator |
| CN112886165A (en) * | 2021-02-26 | 2021-06-01 | 中国电子科技集团公司第九研究所 | Semi-open cavity type microstrip circulator/isolator assembly |
Non-Patent Citations (3)
| Title |
|---|
| YANG YU等: ""Nonlinear Characterization of the MRE Isolator using Binary-Coded Discrete CSO and ELM"", 《INTERNATIONAL JOURNAL OF STRUCTURAL STABILITY AND DYNAMICS》 * |
| 张元庆等: "基于贝叶斯法则的空间自相关误差自相关模型变量选择研究", 《数理统计与管理》 * |
| 闫欢等: ""大功率高相位稳定性P波段铁氧体环行器研制"", 《磁性材料及器件》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113221391B (en) | 2021-09-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112462312B (en) | Automatic calibration method for chip testing machine and application thereof | |
| CN107884626A (en) | A kind of S parameter automatic measurement method based on fissipation factor | |
| CN102520258B (en) | Method for correcting high-frequency characteristic error of electronic component | |
| CN113221391B (en) | Method for guiding material size parameter screening by analyzing circulator performance by BIC criterion | |
| CN107076822B (en) | Test structure and method for judging de-embedding precision of radio frequency device by using lead-in device | |
| CN111614408B (en) | RF test method of Zigbee module | |
| CN113489559B (en) | Passive intermodulation prediction method for coaxial connector under different communication frequency bands | |
| CN101308945A (en) | Debugging method of cavity band-pass filter | |
| CN117056875B (en) | Cable transmission performance analysis method and device based on test data | |
| CN110441599A (en) | A kind of demodulation method and device, readable storage medium storing program for executing for frequency spectrograph | |
| US20210181102A1 (en) | Calibration method and terminal equipment of terahertz frequency band on-wafer s parameter | |
| CN110763977B (en) | System and method for quantitatively measuring and evaluating precision of noise test system | |
| CN114137389B (en) | Method, device, terminal and storage medium for determining S parameter phase of microwave probe | |
| CN113992280B (en) | Insertion loss calibration device of broadband channel production testing clamp and working method thereof | |
| CN110907570B (en) | Organic matter maturity evaluation method and terminal equipment | |
| CN112350772A (en) | Method for testing waveguide by using insertion loss value | |
| CN113242102B (en) | Radio frequency test method and device, equipment and readable storage medium | |
| CN112903434B (en) | Method for constructing joint rock mass strength size effect statistical model | |
| CN115656651A (en) | Noise coefficient test method for spectrum analyzer | |
| CN115001599B (en) | Method for rapidly testing power difference of same-frequency signals | |
| CN115358648B (en) | Management system and method for standard sample in chip test | |
| CN114114121B (en) | Method for rapidly and automatically calibrating line loss of WIFI static test link | |
| CN116756683A (en) | Method and device for determining stability of mass spectrometer, storage medium and electronic equipment | |
| Shen et al. | Noise figure measurement of narrow-band low noise system | |
| CN116087858A (en) | Ultra-wideband TRL calibration piece and method for calibrating complex dielectric constant test system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |